Regrets Or Gas: We Never Really Know

The romantic saying is that the eyes are the windows to the soul. This may be endearing to those searching for a soulmate, though the divorce rate suggests a coin toss would do just as well, but is it good enough for a judge to decide upon a term of years? Empirical evidence suggests not.

A famous experimental paradigm called “mind in the eyes” purports to demonstrate this ability. You are shown a photograph of a pair of eyes, accompanied by a short list of words that describe a mental state or attitude, such as irritated, sarcastic, worried and friendly. Then you are asked to pick the word that best matches the emotions the eyes express. My lab has confirmed that test subjects perform marvelously at this task, selecting the expected word more than 70 percent of the time on average, based on a study we conducted using over 100 test subjects.

Northeastern psych prof Lisa Feldman Barrett questioned this result.

My lab, however, has also discovered a hitch in this paradigm: If you remove the list of words and ask test subjects to “read” the eyes alone, their performance plummets to about 7 percent on average. The word list, it seems, acts as a cheat sheet that helps test subjects unconsciously narrow down the possibilities. People turn out to be quite bad at inferring emotions without context. This includes judges and juries.

From a distance, we believe in our mad skillz in reading other people from non-verbal cues. But believing we know what they’re really thinking isn’t the same as knowing. When we really know someone, enough to provide context to their cues, we may stand a good chance of reading them well. When we don’t, our chance of correctly imputing thought or emotion to their cues, their words, is slim.

And yet, much of our legal system is predicated upon the belief we can accurately gain insight into people’s heads by scrutinizing their demeanor, We know if they’re lying. We know if they’re sincere. We know if they’re remorseful. But what if that look, those eyes, are really telling us they’ve got gas and are trying desperately to hold it together as they stand before the court, sit on the witness stand.

Knowing a few judges well, there is little doubt that most want to do right. While what constitutes “right” is an open question, most are not the venal, or stupid, or selfish and lazy dictators they’re often seen to be. But judges, like every other person, fall prey to their inflated belief of their ability to “read” people. Some realize this, but also realize there isn’t much they can do about it. Their job is to make credibility determinations, whether at hearings, trials or sentences. That they have no magic skills to do so is a recurring gap in the system, but their job nonetheless.

Perhaps the biggest failing for the wise and well-intended judge is that recognizing these limitations, that they don’t really have a clue who is sincere, trustworthy, truthful, they go with the odds or their gut. They find the cop more credible than the defendant because that’s how they read the odds. They find the defendant with whom they can better relate to be more remorseful, and thus deserving of greater mercy at sentence. Or, to put it less gently, they punt, they succumb to in-group bias, and most importantly, they cover their ass.

After all, if you’re going to make a mistake, why make one that puts your position in jeopardy when the alternative will gain far greater acceptance and place a judge above scrutiny. The Federal Sentencing Guidelines are a perfect example.

For the last decade plus, since the Supreme Court’s 2005 decision in Booker, the guidelines have been advisory rather than mandatory. But most lawyers know the idea of “advisory” sentencing guidelines sounds great, but doesn’t always play out so well. With the opinion in Beckles v. United States, the Supreme Court keeps holding on to the fantasy that the guidelines are just advice, not rules.

Sentence a defendant within the Guidelines and no one will question whether the sentence is reasonable. Depart below and all hell can break loose. Depart and have a defendant commit a heinous crime after release and your face will appear on the cover of The New York Post. And not in a good way. Who needs that?

Except imposing a sentence on a human being, caging him for days, months, years, is a responsibility of monumental proportions. To the person, his family, his children, this destroys lives, futures and any chance for happiness and success. The ramifications are monstrously huge. And it’s based on a fallacy, a foundational assumption that judges have not only the magic power to know what number of years is the right number, but whether any variance within the range should be based on the sincerity of a defendant’s assertion of remorse to the court.

Nathan Burney twitted an MIT tech review post about Artificial Intelligence being used to set bail or release, My slightly snarky gut reaction was yes, but not because AI is such a great thing. Rather, because judges do such an awful job of it, “too chickenshit to do anything other than what the prosecutor asks them.” This is somewhat unfair to say, though not entirely untrue. We ask judges to do something no human being can do, then castigate them for doing it wrong.

Is AI, or more broadly, technology, the better answer? Hardly. It suffers from its own failties, from GIGO to insufferable glitches and failures. And by eliminating empirically sound but politically unacceptable truths from the calculus, it’s just another lie in which we want to believe.

But our expectation that judges and jurors can tell things about people from their demeanor, their testimony, their words, is no better. Science, the real stuff rather than the junk forensics allowed into trial daily because some judge somewhere shrugged and muttered, “whatever, I’ll allow it,” tells us that we do a lousy job of convicting the guilty. And so we continue to fill prison cells based on a series of beliefs that we know to be baseless.

It’s not for lack of desire to do the job right, to do “justice,” whatever that is. It’s that we’ve created a system based on little more than ancient beliefs in voodoo which we now know to be false. Yet, it remains good enough to destroy other people’s lives.

13 thoughts on “Regrets Or Gas: We Never Really Know

  1. Billy Bob

    We do a lousy job of exonerating the innocent as well. As in, “no good deed goes unpunished.” We knew all of the above already, doc, but thanx for spelling it out.

    We’ve been pushing the AI solution in the comments section here several time–somewhat facetiously– but now see what you’re driving at. The real answer might be to starve the beast. If there were less profit in the judicial arena, there might be fewer abuses and misuses of process. Too many players with too much time on their hands will engineer judicial grotesqueries of monstrous proportions. After all, they have “a job to do.” Certainly we crave “law and order”, but not that which is arbitrary and capricious, with no sound basis in fact, procedure or precedent. Nicely done for a cold and windy Sunday in March, irregardless.


    Scott, as I have publicly admitted, I am terrible at judging credibility or assessing an offender’s state of mind.

    Evaluating the credibility of cops vs the accused is a huge problem for me, as I have admitted to you and others as well. In the same vein, and as you well know, I once sentenced a bank robber (Shon Hopwood) to a long term in prison. He got out, earned a law degree, clerked for a well-known federal appellate judge and is now at Georgetown Law’s appellate clinic as a senior fellow. Later, I said my sentencing instincts suck–a self-evident “duh!”.

    Your post is very important. It starkly emphasizes that judges are no better than anyone else at evaluating the inner workings of other human beings. Since somebody has to do it in our legal system, judges will continue to try. But Like a cancer drug that only slows the progression, there is no good antidote for the inability of judges (or juries) to do assess the complexity of other humans.

    The best we can do is to repeat a mantra: Always doubt yourself when judging another human–remember that you will frequently fail and that the consequences of your failure will be dire.

    Thanks for writing this post. We need to be constantly reminded of our disabling weaknesses.

    All the best.


    1. SHG Post author

      What I find myself struggling with, Judge, is that we have a system grounded in an ancient lie, that human beings, even wise and exceptionally sincere ones, can put on a robe and suddenly gain the ability to do something that can’t be done. I don’t write this to harp on the point that judges are only human, but that the absurd Rube Goldberg machine we’re constructed and use to destroy lives doesn’t really work, except in our blind belief in the magical abilities of judges. And yet, we persist in this misadventure by ignoring that this system rests on a firm foundation rather than fixing problems that we know, empirically, exist.

  3. Lisa Feldman Barrett

    Scott, thanks so much for citing my NY Times op-ed on emotion and the law. If your readers are interested, my new book (published last week) has a full chapter that goes deeply into the many disconnects between the legal system and neuroscience findings of the last 10 years. It’s called “How Emotions are Made: The Secret Life of the Brain,” and more info is at

    For example, consider the heat-of-passion defense, that your emotions “took you over” and therefore you’re not as culpable as you would be for a premeditated crime. This distinction is rooted in the belief that emotion and reason are separate in the brain, with the latter controlling the former. However, when we look at the structure of the brain, there are no neurons that are dedicated to emotion or to cognition, so one cannot regulate the other. This is another area where modern neuroscience and the legal system should sync up.

    1. SHG Post author

      Thanks, Lisa. Your op-ed was exceptionally thought-provoking, and very much appreciated. The intersection between neuroscience and law is quite significant, but I fear law is so far behind serious science that efforts to introduce it, whether to the judge or jury, remains decades (if ever) away. We have so much to learn before we can move away from our mindless reliance on ancient voodoo.



    I am curious about our failure at “fixing problems that we know, empirically, exist.” As I read these words, I find an implication that we know what the fixes are but refuse to implement them. If I read those words correctly, I wonder what those fixes might be. But, perhaps that is a question to be addressed in another post.

    All the best.


    1. SHG Post author

      Or 100 posts. Junk science (not just the crazy stuff like duct tape analysis, but ballistics, fingerprints and even lab tests), eyewitness testimony, identification procedures, “training and experience,” dog hits, Reid Technique confessions, to name a few. We know better now, yet each persists because precedent.

  5. Keith

    But judges, like every other person, falls prey to their inflated belief of their ability to “read” people. Some realize this, but also realize there isn’t much they can do about it.

    When I asked him about how he determined a witnesses credibility without being able to see them, Judge Casey said that experience as a judge will tell you that anyone claiming to read a witness is bluffing. Witnesses, many of whom have never been inside a courtroom will be nervous and appear to be lying for people unfamiliar with such proceedings. He was adamant that listening to them becomes much more important and he felt that being blind may have helped him in that regard.

    What I remember from the chat was a judge that acknowledged his limitations. Sadly, that seems to be a recognition that takes a while to develop, if at all.

    1. SHG Post author

      I remember the first time I was before Judge Casey, and he came out with his dog. It was disconcerting. I think most judges will, in private, concede their limitations. The problem is they still have a job to do. Or as Judge Denny Chin said, “rule and roll.”

      1. David Stretton

        Great. Now I’ve got Alice’s Restaurant playing in my head. I’d do a linky thing but…

Comments are closed.